KR100267123B1 - Manufacturing method for planar optical waveguides - Google Patents

Abstract

The present invention relates to a flat optical waveguide that extends flat optical preforms to produce flat optical glass rods with substantially smaller cross-sectional dimensions than conventional preforms and achieve the optical circuit pattern using lithographic techniques.

The optical preform may be inserted into a groove of the substrate to prepare a flat optical preform.

Description

Manufacturing method of flat optical waveguide

1 is a schematic of a substrate with a handle attached for support during the present process.

2 to 8 are cross-sectional views of flat optical waveguides at various stages in the manufacturing method according to the invention.

9 shows an example of an optical circuit pattern in the form of a flat optical waveguide manufactured by the present invention.

10 (a) and 10 (b) show another embodiment of the present invention.

11 (a), 11 (b), 11 (c), 12 and 13 show yet another embodiment of the present invention.

14 (a) and 14 (b) show another embodiment of the present invention.

FIG. 15 is a schematic view of disposing an optical fiber and a cross coupling circuit containing cores and clads on a substrate.

16 (a) and 16 (b) are cross-sectional views of two optical fibers along the line A-A of FIG.

[Background]

FIELD OF THE INVENTION The present invention relates to a method for producing hybrids of planar optical waveguides, in particular flat optical waveguides.

Flat waveguides have been used as passive components in optical inter-connection systems. Such a waveguide is distinguished from a cylindrical dielectric waveguide such as an optical fiber in that the waveguide is substantially rectangular in cross section. Conventional methods of making such waveguides are generally uneconomical and require strict manufacturing oversight, with the resulting waveguides having relatively high optical losses as compared to optical fibers.

Conventional methods of making flat optical waveguides employ substrates having a first refractive index and having preselected final dimensions of the formed flat optical waveguide. Materials having a second refractive index different from the refractive index of the substrate are applied to the substrate using a variety of methods including standard soot deposition techniques well known in the art. (Reference: US Pat. Nos. 3,806,223 and 3,934,061 to Keck et al.) The predetermined refractive index differences are achieved by using one or more of the following doped silicas.

Titanium oxide, tantalum oxide, tin oxide, niobium oxide, zirconium oxide, aluminum oxide, lanthanum oxide, germanium oxide (germanium oxide) or other suitable refractive index conversion dopant material.

An optical circuit inside the flat waveguide is typically formed by a lithographic process similar to that used in the manufacture of semiconductor devices, as described in US Pat. No. 4,425,146 to Izawa et al. Another prior art method is described in U.S. Patent No. 3,873,339 (Hudson), wherein the focused laser beam is used only to melt the material forming part of the preselected optical circuit, and the remaining melt does not melt. Unused material is removed by cleaning or etching methods. The use of the slab technique is widely used in the manufacture of semiconductor devices. The above techniques are useful because detailed patterns (in the case of the invention, optical circuit patterns) can be produced.

The lithography process starts with a structure that contains the materials needed to make a preferred electrical or optical circuit. The structure is coated with a photo-resistive material. The photosensitive material is exposed to light using a mask that selectively exposes a portion of the photosensitive material. The mask is in the shape of a preferred circuit pattern. The exposed photosensitive material is developed in a developing solution designed for the type of photosensitive material used. The base structure is then etched, for example by reactive ion etching, so that the mask pattern is transferred to the base structure.

In the case of manufacturing flat optical waveguides, an alloying material, such as chromium, is usually coated on the base structure before applying the photosensitive material. The chromium layer is necessary because the photosensitive material alone cannot withstand the etching conditions required to etch the optical circuit into the underlying glass structure. The photosensitive material is exposed and developed as described above, and the optical pattern is transferred to an intermediate chromium layer using a chrome etching solution. Thereafter, the optical pattern is transferred to the base glass structure using, for example, reactive ion etching.

Each of these conventional etching methods involves applying a very thin layer to form the core region of the waveguide. The core region guides most of the light along the waveguide. Small perturbations in the manufacturing process can result in non-uniform core structures with very high optical losses, in particular relatively high optical losses compared to optical fibers attached to flat optical waveguides. Therefore, in the conventional method, a strict control on the deposition process is required to obtain the thickness of the predetermined core area. In particular, there is a further need in the manufacture of flat optical waveguides for use in single-mode systems using fibers having a core diameter of 10 μm or less.

Problems inherent in the conventional method of manufacturing flat optical waveguides are as follows.

1. The light loss is relatively high compared to the loss of optical fiber.

2. Expensive manufacturing supervision is necessary to keep the light loss to a minimum.

3. Design and geometries are limited.

It is therefore an object of the present invention to provide a flat optical waveguide with lower light loss than that produced by conventional methods. Another object of the present invention is to provide a more economical process than before by combining improved optical performance and mass production capacity through the use of lithographic technology. It is still another object of the present invention to provide a flat optical waveguide having various types of core layers and various refractive index profiles for use in various fields.

[Summary of invention]

The above and other objectives are to form a flat molten glass structure of glass having at least two refractive indices, and then to reduce the thickness of the structure to produce a planar optical cane of a predetermined final dimension. Is achieved by a process for producing a glass blank used for optical waveguide manufacture.

Another object of the present invention is to form a flat structure by combining a glass substrate having a first refractive index and a second glass region having a different refractive index to form a flat structure having a predetermined final dimension. Heating and stretching, removing material from the flat optical glass rod using lithographic technology to produce a preselected optical circuit thereon, and combining the glass rod with additional material regions on at least one region from which the material has been removed. It is to provide a process for manufacturing a flat optical waveguide consisting of a step. In one embodiment of the invention, the flat structure is formed by placing at least one optical fiber preform within at least one slot in the glass substrate.

The "planar optical cane" refers to a structure manufactured by stretching a solidified body having a predetermined refractive index profile, wherein the cross-sectional diameter of the solidified body is reduced, and The preselected refractive index file of the solidified body is present in proportion to the flat light glass rod after it is stretched.

Detailed description of the invention

The accompanying drawings are not intended to indicate the size or relative proportions of the elements.

The present invention uses a hybrid process to produce flat optical waveguides. The process starts with the substrate as shown in FIG. It is essential that the substrate has substantially larger dimensions than desired for the final flat optical waveguide. The material of the substrate is selected to match the thermal and mechanical properties of the material used as waveguide conductors and films. Typically, it will be necessary for the substrate to be made of silica. However, it is possible to use borosilicate or soda lime glass as the material of the substrate as a waveguide conductive material comprising some fluorine composition. Handles 2 and 3 may be attached to facilitate steering during the manufacturing process.

In the process, the next step is to apply one or more layers of materials to the substrate surface having a refractive index that is different from the refractive index of the substrate. Typically, the predetermined refractive index difference is achieved using one or more of the following doped silicas; Titanium oxide, tantalum oxide, tin oxide, niobium oxide, zirconium oxide, aluminum oxide, lanthanum oxide, oxide Germanium oxide, fluorine or other suitable refractive index conversion dopant material.

Dopants for other purposes may be used, for example, erbium or neodymium may be used for the amplification of optical signals. In addition, other compositions such as fluorine glass can be used, and the substrate can be formed from Pyrex glass, soda-lime glass, and the like, which is compatible with the thermal and mechanical properties of the optical waveguide conductive material. The soot may be applied using standard soot production techniques and may be applied to all sides only on one side or by rotating the substrate. The material layer so applied preferably consists of a boundary layer 4, a core layer 5 and a cladding layer 6, as shown in FIG. 2.

Other techniques may be used to apply the material layer on the substrate, for example plasma-enhanced chemical vapor deposition (CVD), sol-gel, low pressure CVD or sputtering, and the like. .

The use of the boundary layer 4 depends on the refractive index and the loss characteristics of the substrate 1. If the difference in the refractive index between the substrate 1 and the core layer 5 is too small, the material of the boundary layer 4 is the core of most of the light incident on the flat waveguide resulting in the difference in refractive index between the boundary layer and the core layer. It is selected to a degree sufficient to pass along layer 5. The refractive index of the cladding layer 6 is also selected to allow effective waveguide propagation through the core layer 5.

Another method of the application step is using a sol-gel or slurry casting techniques, as shown in FIG. 10 (a), a dimensional groove 10 (excavated into the surface of the substrate 1) ( The material 11 applied to the dimensional slots is positioned. A dimension groove is to dig a surface of a substrate using, for example, lithography or a dice saw, depending on the size of the arc.

After applying the material, the structure is heated in a furnace to melt the refractive index generating material and provide a flat photoform. The melting process should preferably take place quickly to reduce the diffusion of dopants in the various soot layers. If dehydrate of the soot layer is desired, the melting step can be carried out in a chlorine atmosphere. One example of such a dehydration process is described in detail in US Pat. No. 4,165,223 to D. R. Rowers. The molten structure is heated to a softening point and stretched to produce a flat glass rod of predetermined final dimensions. Typically included reduction ratios are 50: 1 or less, preferably in the range of 10: 1 to 20: 1. The softening point and aspect ratio (width: height) of the molten structure should be chosen to avoid geometric distortion during the attenuation step. As shown in Figure 3, by using rounded corners instead of sharp shapes, geometric distortion can be reduced.

A preselected optical circuit is fabricated using the slab technique on the flat optical glass rod. As shown in FIG. 4, a metal or alloy coating material 7 and an organic photosensitive coating 8 are applied to the flat light glass rod. Thereafter, the master mask is aligned on the flat light glass rod and the master mask pattern is transferred to the organic photosensitive coating by conventional photo-slab techniques. The exposed organic photosensitive coating is removed by washing the flat glass bar in the developer, and the alloy coating in the exposed area is removed using a commercially available chromium etchant. After removing this coating, the only coating remaining on the flat light glass rod is in the pattern of the master mask, as shown in FIG. Any remaining organic photosensitive material is washed off with acetone. For example, as shown in FIG. 6, the pattern is transferred onto a flat light glass rod by a reactive ion etching method.

In one embodiment of the present invention, as shown in FIG. 9, after completing the etching, the unwanted portion of the second glass is removed, and the relatively wide portion of the unetched glass rod is removed from the flat waveguide. The etching step is performed to remain at the lateral edges. In a preferred embodiment of the present invention, etching removes 15-30 μm wide trenches 33 in the vicinity of waveguide paths 15, 16 and 17. This portion helps to protect the preselected optical circuit from physical damage during subsequent processing.

Any remaining alloy coating is removed using a commercial chrome etchant. For example, as shown in FIG. 7, the optical waveguide of the overclad layer 9 can be flattened using conventional soot deposition techniques or other thin film techniques such as plasma enhanced CVD, sol-gel, low pressure CVD or sputtering. Applies to

In a preferred embodiment, the substrate is a fused silica slab with initial dimensions of 1/2 inch thickness, 2 inches wide, 14 inches long with a refractive index of 1.458. The substrate is shaped and shaped into rectangular cross sections using conventional glass grinding techniques. For example, handles 2 and 3 made of T08 (pulverized silica in use) rod melt the handle to the substrate and adhere to the substrate under flame. The handle allows the substrate to be placed on a glass-working lathe.

The glass working shelf is equipped with a burner to J.F. Hyde, US Pat. No. 2,272,342 and M.E. A flame hydrolysis / oxidation process similar to that described in Nordberg's US Pat. No. 2,326,059 is performed. Conventional vaporizers or fountain bubbler devices are used to transfer chemical reactants to the burners. See US Pat. No. 4,314,837 to Blankenship and US Pat. No. 3,826,560 to Schultz. Similar to that described in US Pat. No. 3,698,936, a discussion of the temperature characteristics of the flame produced by the burner is described in M. Elder and D. Powers' “Benefits of Light Waveguide Sparks,” Atlanta Georgia, page 1986, page 74 Technical Summary of the 1986 Conference on Fiber Optic Communications ”.

The boundary layer 4 shown in FIGS. 2 to 7 is not necessary because the structure in which the substrate is molten is silica and has the necessary refractive index with respect to the refractive index of the core layer 5. The cross section of the glass rod without the boundary layer used in this example is shown in FIG. Reference is made to FIGS. 2 through 7 describing the slab process used in this example for convenience only, since FIGS. 2 through 7 show a boundary layer not used in this example. A core layer 5 of about 100 μm thickness having a refractive index of about 1.464 and composed of SiO ₂ and 8 mass% GeO ₂ is applied to the substrate. Thereafter, a cladding layer 6 of pure silica soot about 100 μm thick is applied over the core layer.

The resulting structure is placed in a furnace for about 20 minutes at a temperature of about 1540 ° C. to melt the core and the clad layers.

The molten structure is placed in a vertical furnace and heated to about 2100 ° C. A second furnace is equipped with a gripping and pulling machine to stretch the molten structure. The molten structure is lowered to a high temperature region that raises the temperature of the molten structure to a softening point. The tensioning mechanism stretches the molten structure by pulling the bottom of the structure out of the high temperature region of the furnace at a rate faster than the rate at which the molten structure is lowered into the hot region. The molten width and thickness thereof decrease while its length increases and elongates. Thus, the flat light glass rod is fabricated about 0.16 inches wide, 0.04 inches thick and 30 inches long. The resulting thickness of the core layer of glass is 6-8 μm. In yet another embodiment, the core layer may be 8-9 μm. The number of each flat optical device that can be manufactured from one flat light glass rod depends on the type of device being manufactured. For example, as shown in FIG. 9, a 3 microliter splitter is about 1 inch long so that one flat light glass rod having an elongated width that matches the width of the device can produce about 30 devices. Can be.

The flat optical glass rod is washed repeatedly in a solution of deionized water, acetone and 1-2% HF. A chromium coating (approximately 2000 microns thick), such as the Chrome Target manufactured by the Materials Research Corporation in Orangeburg, 10962, New York, was applied to flat glass rods using RF-sputtering technology. Apply. An organic photosensitive coating 8, such as S1400-17, manufactured by the Shipley Company, Newton, Mass., Is then spin coated onto the chromium surface at 3000 rpm. The coated flat glass bar is baked in an oven at 110 ° C. for 20 minutes.

Using conventional techniques, the master optical circuit mask is manufactured in a preselected optical circuit pattern. An example of the optical circuit pattern is shown in FIG. The light pattern of the above example becomes a well known 3 kHz splitter device. Light enters the device through the inlet 15. Some of the light exits to exit 16 and some of it exits to exit 17.

In one embodiment of the invention, the coated flat glass pane is fed to a slab machine. The machine aligns the glass rod with a master optical circuit mask and exposes the organic photosensitive coating to ultraviolet light. The preselected optical circuit pattern is thereby transferred to the organic photosensitive coating. The pattern is developed in an organic photosensitive coating using a photosensitive developer such as the Microposit 352 developer manufactured by the Shipley Company, Newton, Massachusetts. The coated glass rod is rinsed in deionized water and dried. In addition, the exposed positive organic photosensitive coating is removed during this step.

The chromium coating in the exposed area of the flat glass glass rod is removed using a commercial chrome etch solution such as Chrome Etch manufactured by KTI Chemicals, Sunnyvale, California. Thereafter, the remaining organic photosensitive coating is removed by washing the flat light glass rod in acetone, rinsing in deionized water and drying. As a result, the flat optical glass rod has a chromium coating in the pattern of the preselected optical circuit.

The unprotected glass region of the flat light glass rod is etched using reactive ion technology. The remaining chromium coating is removed using a commercial chromium etching solution. The flat light glass rod is scrubbed in deionized water, a commercial glass cleaner and 1-2% HF solution, rinsed and dried in deionized water.

Thereafter, an overclad layer 9 (FIG. 7) of at least about 15 [mu] m is applied over the optical circuit by conventional soot deposition techniques. If passive alignment is required for pigtail arrays, an overclad layer 9 of about 62.5 μm should be applied. The overclad layer 9 is silica doped with about 8% by mass of B ₂ O ₂ to reduce the melting temperature and doped with about 1% by mass of GeO ₂ to have a refractive index of about 1.458. The dopant levels should be adjusted accordingly to produce waveguides different from single mode operation at 1.3 to 1.55 μm. The cladding material melts for about 20 minutes at a temperature of about 1320 ° C. to become a flat optical waveguide and covers the optical circuit without leaving any gaps in the clad layer.

The flat optical waveguide made by the process of the present invention shows improved optical performance. Attenuations including coupling losses induced during the measurement were measured to be less than 0.02 dB / cm. Although the theoretical coupling loss is due to the measuring device, the attenuation calculated in the flat optical waveguide produced in the process of the present invention is less than 0.01 dB / cm. This is compared with 0.05-0.1 dl / cm by the conventional technical process. The substantial reduction in this reduction is believed to be the result of the smoothing and reduction of the defects during redraw.

Another embodiment of the invention combines one or more flat optics in an optical circuit pattern. Another alternative embodiment is a process for a series of flat optical devices by successively exposing a portion of a coated flat glass glass rod using a master slab pattern.

Another embodiment is a process for producing a flat light glass rod having a longer length by injecting a glass rod into a device that continuously exposes an area of the glass rod to a predetermined optical circuit master mask. This is illustrated in Figure 14 (a), in which the long flat glass bar 22 coated with the chrome coating 7 and the organic photosensitive coating 8 exposes a continuous area of the long flat light glass bar 22. In order to arrange on the mask mask 23. The exposed long flat light glass rod 22 is etched as described above and cut into each flat optical waveguide. Alternatively, as shown in FIG. 14 (b), a plurality of master masks 24, each made of specific optical circuit patterns 24a, 24b, and 24c, are moved as the long flat light glass rod is moved to the exposed position. It may be indexed to a position on the long flat light glass rod 22. In this way, one long flat light glass rod 22 can be used to manufacture several different types of flat light waveguides.

On the other hand, another method of forming the preselected refractive index profile is to etch the grooves 10 of the correct dimensions to match the preselected optical circuit pattern on the unstretched substrate 1, as shown in FIG. As described above, the groove 11 is filled with the material 11 using soot deposition, sol-gel or slurry casting techniques. The refractive index of the material 11 is different from the refractive index of the substrate. The cross-connect layer 12 may be applied using the soot deposition technique described above. The resulting structure is melted and stretched as described above. The molten structure is etched as described above because it is needed to further clarify the preselected optical circuit pattern. In particular, as shown in FIG. 10 (b), it is possible to remove the region of the crosslinking layer 12, leaving the crosslinking passage 13.

Another method etches the correct dimensional grooves 14 in the substrate 1 in accordance with a preselected optical circuit pattern, as shown in FIG. 11 (a). Then large with a core region 16, 16 'having at least one shape (e.g., circular, square, elliptical or D-shaped) or a desired refractive index profile (e.g., step or grade). The core optical fiber 15 (hereinafter referred to as optical fiber preform 15) is positioned in at least one of the grooves 14. The optical fiber preform may optionally consist of core only. In addition, stress inducing materials or members may be included to provide stress birefringence.

In FIG. 11 (a), the optical fiber preform 15 is crushed to expose the core region 16. The optical fiber preform 15 is positioned above the substrate such that its optical axis is parallel to the elongation axis of the substrate.

In another embodiment, the optical fiber preform can be recoated by placing it on a flat substrate without grooves. Alignment projections or elongated grooves may be included in the glass rod to help position the fibers. The shape of the optical fiber preform can be selected based on the expected change during stretching. For example, a circular core may be transformed into an elliptical core. Morphological changes can be controlled to some extent by limiting the soot thickness and by using shaped blanks with stiff cladding.

An appropriate refractive index crosslinking layer 17 is placed over the optical fiber preform 15 and melted as described above. The crosslinking layer 17 can be brought into contact with the surface of the optical fiber preform 15 as shown in FIG. 11 (a), and in advance above the optical fiber preform 15 as shown in FIG. 11 (b). Can be located away at a selected distance. The protective overclad layer 18 may be applied over the crosslinked layer 17 as shown in Figure 11 (c) using soot deposition, sol-gel or slurry casting techniques. The protective layer reduces contamination and / or diffusion of the dopant material during consolidation.

The resulting structure is melted as described above. The molten structure is stretched and etched as described above to more clearly define the preselected optical circuit pattern. As shown in FIG. 11 (c), the alignment groove 25 can be used to arrange the master mask more accurately than the embedded glass rod or optical fiber for proper crosslinking. An alignment protrusion may be used instead of the groove 25.

An example of a simple branched crosslinking layer is shown in FIG. 12, where the crosslinking layer 17 of FIG. 11 (b) is used to leave the crossover optical circuit 19 between the waveguide cores 16 and 16 'after the stretching process. By etching the branch circuit 19 is formed. Another way to form crosslinks between the waveguide conductors 46 and 46'embedded in the substrate is to etch the crosslink passage 20 shown in FIG. Thereafter, the passage is filled with a material 21 having a suitable refractive index for the optical interbonding required using soot deposition, sol-gel or slurry casting techniques. In the embodiments of FIGS. 12 and 13, the waveguide conductor and crosslink circuit are recoated with glass and form a solid waveguide structure.

In another embodiment, a flat light glass rod comprising the core layer (after extension) or a core layer plus a predetermined thickness of the cladding layer is etched as indicated in FIG. 15 to provide a crosslinking pattern to the core layer. do. The crosslinking pattern is raised about 8 microns from the surface of the substrate. The optical fiber with the core 36 and the cladding 37 is located on the side of the core that contacts the raised crosslink circuit 39. Cross-sectional views of two optical fibers along line A-A in FIG. 15 are shown in FIGS. 16 (a) and 16 (b). The alignment may be easy from the alignment protrusion 35 formed in the glass rod. Alternatively, the alignment grooves can be used to match the corresponding protrusions in the means for positioning the fibers. The optical fiber is permanently fixed with low index epoxy or plasma-enhanced CVD to remain in the elevated crosslink circuit. Thereafter, the glass rods and fiber assemblies are recoated with glass using conventional means to form solid waveguide structures having pigtails. By placing the optical fiber in the structure after the stretching process, the fiber can be used for the attachment of the pigtail by pigtail or splicing.

The invention has been described with reference to particularly preferred embodiments. However, one of ordinary skill in the art will be able to make various changes and modifications without departing from the spirit and scope of the invention as defined by the following claims. For example, although the present invention has been primarily described for single mode waveguide structures, it can be applied to multimode waveguide structures with moderate variations in dopant degree and dimensions.

Claims (25)

(a) combining a second glass region having a second index of refraction different from the first index of refraction using a soot deposition technique to a glass substrate having a first index of refraction to form a flat structure; (b) heating and stretching the structure to a softening point to produce a flat optical glass rod of predetermined final dimensions; (c) using a lithographic technique to remove metal or alloy coating material and organic photosensitive coating from a portion of the flat optical glass rod to form a predetermined optical circuit thereon; And (d) using a soot deposition technique or other chemical vapor deposition technique to combine a portion of the overclad material over the region from which the material has been removed with the optical glass rod. . (a) combining a second glass having a second index of refraction different from the first index of refraction using a soot deposition technique into the groove of the dimension of the glass substrate having the first index of refraction to form a flat structure; (b) heating and stretching the structure to a softening point to produce a flat optical glass rod of predetermined final dimensions; (c) using a lithographic technique to remove metal or alloy coatings and organic photosensitive coatings from a portion of the flat optical glass rod to form a predetermined optical circuit thereon; And (d) combining a portion of the overclad material over the area from which the material has been removed with the light glass rod using soot deposition or other chemical vapor deposition techniques. 3. The method of claim 2, wherein said second glass consists of an optical fiber preform positioned in a groove of said glass substrate. 4. A cross-linking layer is applied after positioning said optical fiber preform in said groove, and said removing step comprises removing areas of said cross-linking layer to leave a cross-linking pattern. Method of manufacturing a flat optical waveguide. 4. The method of claim 3, wherein the optical fiber preform consists of a large core optical fiber located in a groove of the glass substrate. 6. The method of claim 5, wherein a cross-linking layer is applied after placing the optical fiber preform in the groove, and the removing step comprises removing the area of the cross-linking layer to leave a cross-coupling pattern. Method of manufacturing an optical waveguide. The method of claim 1, wherein the aspect ratio of the flat structure is selected to prevent geometric distortion during the stretching step. 2. The method of claim 1, wherein the temperature at which the flat structure is stretched is selected to prevent geometric twisting during the stretching step. The method of claim 1, wherein the edges of the flat structure are rounded to reduce geometric torsion during the stretching step. The method of claim 1, wherein the grinding ratio of cross-sectional dimensions during the stretching step is in the range of 10: 1 to 20: 1. The method of claim 1, wherein the second glass is initially composed of a layer having a thickness of 100 µm or more, and the layer is reduced to a thickness of 6 to 8 µm during the stretching step. . 2. The method of claim 1, further comprising melting the layer of overclad material, wherein the composition of the overclad layer is carefully selected to cover the area from which the material is removed after the overclad layer is melted. Method for producing a flat optical waveguide, characterized in that it further comprises. The method of claim 1, further comprising cutting the flat optical glass rod into a plurality of pieces prior to the removing step, wherein the pieces consist of an area of the flat optical glass rod. . The method of claim 1, further comprising cutting the flat optical glass rod into a plurality of pieces after the removing step. 2. The method of claim 1, wherein the step of combining the overclad material comprises positioning the optical fiber in contact with the preselected optical circuit and in contact with the glass rod. 16. The method of claim 15, wherein said removing step further comprises an alignment means for assisting the optical fiber in the proper position. 16. The method of claim 15, further comprising overcoating the set of optical fibers and glass rod portions with glass. 3. The method of claim 2, wherein the aspect ratio of the flat structure is selected to prevent geometric torsion during the stretching step. 3. The method of claim 2, wherein the temperature at which the flat structure is stretched is selected to avoid geometric twisting during the stretching step. 3. The method of claim 2, wherein the edges of the flat structure are rounded to reduce geometric torsion during the stretching step. The method of claim 2, wherein the attenuation ratio of the cross-sectional dimension during the stretching step is in the range of 10: 1 to 20: 1. 3. The method of claim 2, further comprising melting the layer of overclad material, wherein the composition of the overclad layer is carefully selected to tightly cover the region from which the material has been removed after the overclad layer is melted. Method for producing a flat optical waveguide, characterized in that it further comprises. The method of claim 2, wherein the step of joining the regions of the overclad material comprises positioning an optical fiber connected to the preselected optical circuit to contact the optical glass rod. . 24. The method of claim 23, wherein said removing step further comprises alignment means for helping said optical fiber to be in a proper position. 24. The method of claim 23, further comprising overcoating the set of optical fibers and the light glass rod region with glass.